Inverse lithography and machine learning for mask synthesis
Techniques relating to synthesizing masks for use in manufacturing a semiconductor device are disclosed. These techniques include providing, by a processor, a design pattern for a semiconductor device as input to a trained machine learning (ML) model. The techniques further include performing, using the ML Model, a plurality of dilated convolutions relating to the design pattern, and inferring, using the ML model, one or more masks for use in manufacturing the semiconductor device, based on the plurality of dilated convolutions.
1. A method comprising:
converting, by a processor, a design pattern for a semiconductor device to a pixel domain, to provide a converted design pattern;
providing the converted design pattern as input to a trained machine learning (ML) model;
performing, using the ML Model, a plurality of dilated convolutions relating to the converted design pattern;
inferring, using the ML model, a mask for use in manufacturing the semiconductor device, based on the plurality of dilated convolutions; and
converting the inferred mask from a level-set function to a polygon representation, wherein the level-set function defines contours of the inferred mask.
2. The method of claim 1 , wherein each convolution in the ML model is a dilated convolution comprising a kernel dilation factor greater than 1.
3. The method of claim 1 , further comprising:
determining a kernel dilation factor for each of the plurality of dilated convolutions based on a target ambit.
4. The method of claim 3 , further comprising:
determining depth and kernel width for the ML model based on the target ambit.
5. The method of claim 1 , further comprising providing a simulated image relating to using the design pattern as a mask for manufacturing the semiconductor device, to the ML model.
6. The method of claim 1 , wherein the ML model comprises one or more of a mirrored convolution and a rotated convolution.
7. The method of claim 1 , wherein the ML model is trained in part based on the design pattern for the semiconductor device.
8. A system comprising:
a processor; and
a memory storing instructions, which when executed by the processor, cause the processor to perform operations comprising:
converting a design pattern for a semiconductor device to a pixel domain, to provide a converted design pattern;
providing the converted design pattern as input to a trained machine learning (ML) model;
performing, using the ML Model, a plurality of dilated convolutions relating to the converted design pattern;
inferring, using the ML model, one or more masks for use in manufacturing the semiconductor device, based on the plurality of dilated convolutions; and
converting the inferred mask from a level-set function to a polygon representation.
9. The system of claim 8 , wherein each convolution in the ML model is a dilated convolution comprising a kernel dilation factor greater than 1.
10. The system of claim 8 , the operations further comprising:
determining a kernel dilation factor for each of the plurality of dilated convolutions based on a target ambit.
11. The system of claim 10 , the operations further comprising:
determining depth and kernel width for the ML model based on the target ambit.
12. The system of claim 8 , wherein the operations further comprise providing a simulated image relating to using the design pattern as a mask for manufacturing the semiconductor device, to the ML model.
13. The system of claim 8 , wherein the ML model comprises one or more of a mirrored convolution and a rotated convolution.
14. The system of claim 8 , wherein the ML model is trained in part based on the design pattern for the semiconductor device.
15. A non-transitory computer program product comprising:
a non-transitory computer-readable storage medium having computer-readable program code embodied therewith, the computer-readable program code executable by one or more computer processors to perform an operations, the operations comprising:
converting a design pattern for a semiconductor device to a pixel domain, to provide a converted design pattern;
providing the converted design pattern as input to a trained machine learning (ML) model;
performing, using the ML Model, a plurality of dilated convolutions relating to the design pattern;
inferring, using the ML model, one or more masks for use in manufacturing the semiconductor device, based on the plurality of dilated convolutions; and
converting the inferred mask from a level-set function to a polygon representation.
16. The non-transitory computer program product of claim 15 , wherein each convolution in the ML model is a dilated convolution comprising a kernel dilation factor greater than 1.
17. The non-transitory computer program product of claim 15 , the operations further comprising:
determining a kernel dilation factor for each of the plurality of dilated convolutions based on a target ambit.
18. The non-transitory computer program product of claim 17 , the operations further comprising:
determining depth and kernel width for the ML model based on the target ambit.
19. The non-transitory computer program product of claim 15 , wherein the ML model comprises one or more of a mirrored convolution and a rotated convolution.
20. The non-transitory computer program product of claim 15 , wherein the ML model is trained in part based on the design pattern for the semiconductor device.